7 Interaction between a Wildfire and the Sea-Breeze
Front
Deborah E. Hanley
AWS Truepower, LLC, 463 New Karner Road, Albany, NY 12205, USA
Email: [email protected]
Phone: 51 8-213-0044 X 1027
Philip Omningham
Earth and Environmental Sciences Division, Los Alamos National Laboratory,
Los A lamos, New Mexico 87545, USA
Email: [email protected]
Phone: 505-667-5092
Scott L. Goodrick
Center for Forest Disturbance Science, USDA Forest Service,
Athens, Georgia 30602, USA
Email: [email protected]
Phone: 706-559-4237
Abstract Florida experiences sea breezes, lake breezes, and bay breezes
almost every day during the year, and there are frequently complex interactions
between many of these breezes. Given the often-rapid changes in temperature,
humidity and wind speed that accompany these breezes, most wildfires and
prescribed fires in Florida are affected in some way by their interaction with
these circulations. In this paper, we explore the interaction between sea breezes
and wildland fires from both an observational and an idealized modeling
perspective. The progression of the sea-breeze front and its interaction with
the smoke plume from a 26,000 acre wildfire are tracked using a variety of
data sources including surface and upper-air o bservations as well as NEXRAD
radar imagery. Idealized numerical simulations of a thermally buoyant plume
interacting with a density current are performed in an effort to enhance
our understanding of the dynamics of the interaction between sea breeze
circulations and the convective column of the fire. Our observational analysis
and idealized modeling results suggest that the arrival of a sea breeze
front induces a temporary, but significant, increase in fire intensity. This
intensification precedes the arrival of the sea-breeze front at the location of
the fire, such that the fire intensity is at a maximum at the time of, and slightly
after, the passage of the front, and decreases gradually thereafter as cooler
and moister air behind the front arrive.
Keywords
Sea-breeze, fire behavior, fire weather, numerical modeling
Remote Sensing and Modeling Applications to Wildland Fires
7.1
Introduction
Weather is one of the most important factors leading to extreme wildfire behavior.
Rapid changes in wind can quickly affect the rate and direction of fire spread,
and may have a significant impact on fire intensity. Changes in temperature and
humidity also may cause unexpected changes in fire behavior. Florida coastlines
experience sea-breezes, lake breezes, and bay breezes almost every day during
the year, and there are frequently complex interactions between many of these
breezes. Given the often-rapid changes in temperature, humidity and wind speed
that accompany these breezes, most wildfires in Florida arc affected in some way
by their interaction with these circulations. Nevertheless, despite the importance
of sea-breeze circulations for fire behavior, the nature of the interaction between
sea-breezes and wildfires remains relatively poorly understood.
Previous studies of sea-breezes depended upon a network of observation sites
in the area of the sea-breeze (e.g., Atkins and Wakimoto, 1997), or used numerical
models to simulate the sea-breeze (e.g., Pielke, 1974; Sha et al., 1991; Rao et al.,
1999; Colby, 2004). Radar and satellite imagery have been shown to be useful in
tracking the structure, evolution and timing of the movement of sea-breeze fronts
(e.g., Atlas, 1960; Wakimoto and Atkins, 1994; Atkins et al., 1995; Atkins and
Wakimoto, 1997), and have also been used to observe smoke plumes from fires
(e.g., Banta et al., I 992; Rogers and Brown, 1997; Hufford et al., 1998), as well
as from other sources of particulate emissions such as volcanoes, dust storms,
and industrial emissions.
In this Chapter, we explore the interaction between sea-breezes and wildfires
in more detail from both an observational and an idealized modeling perspective.
The progression of the sea-breeze front and its interaction with the smoke plume
from the fire are tracked using a variety of data sources including surface
observations from the remote automated weather station (RAWS) network,
surface and upper-air observations from the national oceanic and atmospheric
administration/national weather service (NOAA/NWS), and imagery from the
Weather Surv~illance Radar-1988 Doppler (WSR-88D) site in Tallahassee, Florida.
Given that the sea-breeze may be represented by a density current, idealized
numerical simulations of a thermally buoyant plume interacting with a density
current are performed in an effort to enhance our understanding of the dynamics
of the interaction between sea-breeze circulations and smoke plumes. Although
these idealized simulations do not allow us to determine directly the impact of the
sea-breeze on fire behavior, which would require a fully coupled atmosphere-fire
model (e.g., Clark et al .. 1996ab, 2004; Linn, 1997; Linn and Cunningham, 2005),
they do provide insight into the interpretation of fire behavior observations, and
allow us to make inferences as to how the fire behavior would likely change in
response to the simulated changes in atmospheric circulations. The combination
of observations, idealized modeling results and inferences on fire behavior lead
82
7
Interaction between a Wildfire and the Sea-Breeze Front
to the development of an initial conceptual model of bow the sea-breeze impacts
\\ildfire behavior.
7.1.1
Sea-Breeze Structure and Characteristics
Simpson (1994) provides a detailed overview of the structure and evolution of
the sea-breeze front derived from observations as well as numerical and
laboratory studies. It is generally observed that the development of the seabreeze is dependent both upon the temperature difference between the land and the
adjacent water surface and upon the strength of any prevailing offshore winds.
Offshore flow can often have a dramatic effect on the depth of the sea- breeze
circulation. The depth of the sea-breeze circulation may be less than 50 m with
winds above this level blowing in the opposite direction. As time progresses, the
depth of the sea breeze may reach 1,000 m or more.
The inland penetration of the sea-breeze can vary from 30 - 300 km depending
upon the location. Typical distances in Florida are about 50 km, although strong
prevailing winds that are in the same direction as the sea-breeze can often result
in much deeper penetration, almost to the opposite coast. Florida also has a unique
geographic design, being surrounded by water on the east and west coast of the
peninsula as well as along the Panhandle coastline. Sea-breezes are observed along
all the coasts as well as along the bays, inlets and large lakes like Lake Okeechobee
m central Florida. These lake and bay breezes interact with the coastal sea-breezes
and result in very complicated wind patterns. Forecasting the winds associated
with the sea-breeze passage is also made more difficult by the presence of convex
and concave coastlines, resulting in additional convergence and divergence effects
(Simpson et al., 1977).
Wakimoto and Atkins ( 1994) and Atkins and Wakimoto ( 1997) showed that
there were significant differences between the sea-breezes observed under offshore
and onshore flow conditions. The offshore flow case exhibited stronger low-level
convergence, larger vertical velocities, and higher radar reflectivity values at the
sea-breeze front, also known as a radar thin line. The sea-breeze thin line and the
kinematic sea-breeze frontal boundary are found to be co-located and easily
identifiable on offshore flow days. Onshore flow days typically have a much weaker
front and the radar thin line is often hard to detect. [n the offshore flow case, the
radar thin line is found to increase in intensity during the day and gradients of
temperature and moisture are strongest on these days.
It is generally accepted that sea-breeze circulations are dynamically similar to
density currents (e.g., Simpson, 1994). which are predominantly horizontal flows
dmen by density differences. Atkins et al. (1995) note that in the case of an
11iTshore flow sea-breeze, there was a kinematic frontal structure similar to that
found in density currents produced in the laboratory. Wakimoto and Atkins ( 1994)
calculated the speed of sea-breeze fronts and found them to be closely matched
83
Remote Sensing and Modeling Applications to Wildland Fires
to the theoretical speeds of comparable density currents. Further study by Atkins
and Wakimoto (1997) showed Froude numbers for offshore and parallel flow
sea-breeze events, as well as for gust fronts, compared well to values calculated
from laboratory density cwrent experiments. This comparison was not as well
dermed on the onshore flow days where the sea-breeze was moving slower than
would be expected using density current theory. It is not clear why this discrepancy
exists for onshore flow sea-breezes, although it is suggested that it may be related
to the difficulty in locating the location of the frontal zone in these cases.
7.1.2
Radar Observations of Smoke Plumes and the Sea-Breeze
Many wildfires occur in remote regions where observations are limited; as a result,
the use of satellite and radar imagery is becoming more prominent in fire detection
and monitoring. It has been suggested that remotely sensed data might also be
useful in fire behavior prediction (Hufford et al. 1998). Both smoke plumes from
fires (e.g., Banta et al., 1992) and sea-breeze fronts (e.g., Wilson et al., 1994) are
clearly visible on radar, especially when the radar is in clear-air mode.
Previous studies of the signals returned from the radar during clear-air mode
operation have suggested that the echoes are most likely that of birds or insects
(Crawford, 1949; Hardy and Katz, 1969; Wilson et al.. 1994). Wilson et al. (1994)
found that the thin line echoes commonly seen in association with sea-breezes
correlated to updraft regions and result from insects actively flying downward to
avoid being carried to colder regions of the atmosphere. Atlas (1960) discounts
the idea of echoes resulting from birds or insects since they would have to fly
along in a very narrow beam, lowering their flight as the beam lowered, or fly in
broad waves normal to the beam with the same speed as the beam. Instead, Atlas
suggests the pattern seen is a result of a sharp increase in the refractive index.
The contrast on the radar is highest when the offshore flow is dry so the contrast
is high, leading to higher refractive discontinuities. This would require a sharp
vertical lapse rate. Battan (1973) suggests both insects and refractive gradients
could be responsible for the radar echoes. Despite some uncertainty regarding
exactly what causes the echoes, the sea-breeze signature is readily apparent on
the radar for the present case and can be used to follow the inland propagation of
the sea-breeze front.
Plumes of smoke have also been observed using radar and satellite imagery in
remote locations such as Alaska (Hufford et al., 1998), where plumes from a
large wildfire were readily apparent on radar. The plumes from smaller fires were
not visible, and it was suggested that the plumes in these cases were concentrated
below the level of the radar beam. High reflectivities (i.e., 20- 25 dBZ) were
noted near the head of the fire, however, and Hufford et al. ( 1998) suggested that
radar imagery can be used to provide information on fire location, intensity and
growth, smoke plumes and fire weather.
84
7
Interaction between a Wildfire and the Sea-Breeze Front
Smoke from a major industrial fire in Montreal, Canada, was observed on three
radars operated by McGill University on 23 May 1996 (Rogers and Brown, 1997).
They suggested that fires produce particulate matter (PM) and create fluctuations
in the refractive index, both of which are potentially detectable by the radar. ln this
case, they concluded that the echoes are the result of scattering by the particles.
Banta et aL (1992) also support the idea that particles in smoke are responsible
for the echoes, with the particles having a flat or needle shape (ash platelets or
pme needles).
7.1.3
Effect of Sea-Breezes on Fires
There are three main factors that affect fire behavior: fuel, weather and topography.
Weather is by far the most variable and arguably the most important of the three.
Wind is a strong driver of the rate and direction of fire spread and can also alter
fire intensity as welL Changes in wind direction can transform a low intensity
backing fire (moving into the wind) or a moderate intensity flanking fire (moving
laterally to the wind) into an intense head fire (moving with the wind) capable of
overrunning fire fighters who thought they were located in a safe area.
Firefighters typically monitor weather conditions regularly for any changes
in wind speed or direction at the scene of the fire using hand-held instruments or
portable weather stations. On days where there is adequate moisture, a line of
approaching clouds and a gust front may indicate the approach of the sea-breeze
front On dry days, there may be no clouds to warn of the impending frontal passage
and the first indication may be the arrival of the gust front. Most rrre personnel
do not have direct access to radar or satellite data while on a fire scene. Local
knowledge of the behavior of the sea breeze is extremely important on both wildfires
and prescribed fires as the onset and structure of the sea breeze are typically
highly regular in Florida in spring and summer months. One case of a springtime
wildfire that is of interest is the East fork fire because of the array of data that
was available during the incident and the opportunity it provided to study the
sea-breeze-wildfire interaction in a novel way.
7.1.4
East Fork Fire
The East fork fire began either late on 4 April 2004, or early on 5 April 2004, in
the Bradwell Bay Wilderness area of the Apalachicola national forest. This area
is located in the Florida Panhandle, just southwest of Tallahassee. It is believed
that the fire was human-caused, either arson or accidentaL The fire ultimately
burned over 26,000 acres of the wilderness area. Fire fighting was made difficult
due to the restriction on the use of heavy equipment in the wilderness area itself.
The fire was considered 90% contained by April 16, 2004.
85
Remote Sensing and Modeling Applications to Wildland Fires
The sea-breeze was a major factor in fighting the East fork fire. Winds shifted
direction on most afternoons, and these wind shifts resulted in varying directions
of fire spread. Firefighters had to be aware of these possible weather changes and
be able to adjust tactics as the weather changed. Conditions during this period were
generally dry, and the daily sea-breeze circulations did not bring any beneficial
rains and for the most part were dry events. Because of the dry conditions, the
WSR-880 radar site in Tallahassee was operating in clear-air mode at the time of
the fire. During clear-air mode periods, the sea-breeze front is often clearly visible
on the radar; moreover, smoke plumes can often be seen if they are close enough
to the radar. In this case, both the sea-breeze and the smoke plume were noted
together and the resulting interaction between the plume and the sea-breeze front
could be observed.
The plan of the pre~enl Chapter i~ a~ follow~: in Lht: following section, the data
and methodology for both the case study and the idealized numerical simulations
arc described, while Section 7.3 presents the results of the case study. In Section
7 .4, idealized numerical simulations are presented Lhat explore the interaction
between a buoyant plume and a density current, and Section 7.5 summarizes the
main results of the present investigation and presents a conceptual model for
sea-breeze- fire interactions.
7.2
7.2.1
Data and Methodology
Case Study
The goal of this study is to provide an improved understanding of the interactions
between wildfires and sea-breeze circulations. The foundation of this understanding
is based upon specific observations of one event that will be augmented by
idealized numerical simulations. The atmospheric component can be adequately
described through a combination of surface observations, upper-air soundings
and radar data (both base reflectivity and radial velocity). Description of the fire
is considerably more problematic, as no direct measurements of the fire are
available, except for daily estimates of area burned. As an alternative, we focus
on the behavior of the smoke plume as observed by the Tallahassee WSR-88D
radar as a surrogate, since Hufford et al. (1998) found high reflectivities to be
associated with more intense burning.
The East Fork Fire burned over a period of approximately two weeks in early
April of 2004; however, this study only focuses on the early stages of the fire.
specifically the afternoon of April 5. The location of the East fork fire and the
four nearby weather stations are shown in Fig. 7.1 . Observations of temperature,
relative humidity, wind speed and direction are used to track the passage of the
sea-breeze front. A high-pass filter was applied to the temperature and relative
86
7
Interaction between a Wildfire a nd the Sea-Breeze Front
humidity time series to remove waves with periods greater than or equal to 24 h.
The high-pass filter was accomplished using a Fourier decomposition of a 96 h
ttme scnes for each station and recomposing the time series with the longer wave
periods excluded.
..-··
~~• Tallahassee
.... ;Bloxham
.
<-'
LIBERTY
LEON
~
WAKULLA
~
JVilma
j
\
;
I
~
\
Sanborn
~~
\-....,_
-·"}..--~
"St. Marks (West)
•.~
"'
FRAJ'.;KLJN
...
Carrabelle "
r----~-------.1
E:nst fork fire
\:Veathcr station location~
M
/
I~~
final
fork
lire perim~'lcr
Count~ line
+
f '\WN "Cather SlaiJOn
+
~ V. \ "cathcr >Calion
1::. R..-\ V. ~ "cathcr ''auon
'
~®I
o Z'
Figure 7.1
"
s
10
1"
2~
Fire perimeters and observation locations
As noted above, the Tallahassee WSR-88D radar was operating in clear-air mode
during this period, which is its most sensitive mode of operation (Crum and Alberty,
1993). The antenna has a slower rotation rate than in precipitation mode that
increases the radar sensitivity, and therefore its ability to sense smaller objects in
the atmosphere. Most of the signal in this mode will be the result of airborne dust
and PM. In this mode, the radar scans five different elevation angles (0.5°- 4.5°,
in I increments) and takes about 10 minutes to complete each scan. In this study
we will focus our attention on the three lowest elevations (0.5 °, 1.5°, and 2.5") of
the base reflectivity and radial velocity fields.
7.2.2
Idealized Numerical Simulations
The sea-breeze has been studied extensively in the past, and many studies have
mvolved the use of numerical models to simulate sea-breeze c irculations, both in
an effort to simulate observed cases and in an idealized context to explore their
structure and dynamics. From the perspective of idealized simulations of the
87
Remote Sensing and Modeling Applications to Wildland Fires
sea-breeze, Dailey and Fovell (1999), Fovell and Dailey (200 I), and FoveJI (2005)
used a 3D cloud-resolving numerical model with horizontal resolutions as high
as 500 m to explore various aspects of the interactions between sea-breeze fronts
and horizontal convective rolls in the boundary layer. Several investigators have
employed even higher resolution 1n 2D configurations to explore the parallel
between sea-breeze fronts and density currents (e.g., Sha et al., 1991), and in this
regard, simulations exploring the dynamics of thunderstorm outflow boundaries
are also relevant (e.g., Droegemeier and Wilhelmson, 1987; Xu et al., 1996; Xue
et al., 1997).
As discussed further below, our approach follows these studies in that the
sea-breeze is idealized as a density current; however, we employ a large-eddy
simulation (LES) model focusing solely on the atmospheric boundary layer, with
model grid spacing on the order of 10m in all three dimensions, to explore the
dynamics of the interaction of this density current with a buoyant plume that is
representative of the smoke plume from a wildland fire. Since it is expected that
the most significant interaction between the sea-breeze and a fire and its attendant
plume will occur primarily with the passage of the sea-breeze front, this approach
appears to be justified, although we acknowledge that there are several details of
observed sea-breeze circulations that cannot be represented by the present model.
The LES model to be used is described by Cunningham et al. (2005), and is
based on the dynamical core of the weather research and forecasting (WRF) model
in physical height coordinates (Skamarock et al., 2001 ). The domain size used in
these simulations is a rectangular box of size 800 x 3,200 x 1,000 m in the x, y,
and z directions, respectively, with a uniform grid spacing of 10m in all three
directions. Boundary conditions are periodic in the x-direction and open-radiative
in they-direction. Other details of the LES model are identical to those described
by Cunningham et al. (2005).
As noted previously, there is significant evidence that a sea-breeze front can be
interpreted as a density current, particularly when the ambient wind is in the
offshore direction. Density currents have been explored extensively by numerical
models, in an atmospheric context primarily in connection with thunderstorm
outflows (e.g., Droegemeier and Wilhelmson, 1987; Xu et al., 1996; Xue et al.,
1997), but also in oceanographic (e.g., Ozgokmen et al., 2004) and general fluid
mechanics (e.g., Hartel et al., 2000) contexts, and even with respect to the backdraft
phenomenon in building fires (e.g., Fleischmann and McGrattan, 1999). Here we
initialize the density current with a cold pool that is uniform in they-direction in
the presence of an opposing flow, to simulate the advance of a sea-breeze front in
the presence of offshore winds.
As a final comment concerning the modeling approach, we emphasize that
the goal of the idealized modeling portion of this study is not to reproduce the
behavior in the observed case, but rather to gain insight into the basic dynamical
processes associated with the interaction between a sea-breeze front and a buoyant
plume representative of those seen with wildland fires.
88
7
lnteractioo between a Wildfire and the Sea-Breeze Front
7.3 Case Study Analysis
The East fork ftre began late on April 4, 2004, or early on April 5, 2004, and is
sullpected to have been started by human causes, either accidental or arson. The
fire burned in timber and southern rough fuel groups. According to the National
Interagency Coordination Center reports, by April 7 the fire reached 7,000 acres
(28.3 km2 ) and was only 20% contained. Spotting of up to 0.25 mile (0.4 km)
ahead of the main fire front was observed with rates of spread of 20 - 30 chains
1
per hour (0.11 - 0.17 m-s- ) reported. This observed fire activity continued and by
April 8. the area burned had increased to over 8,400 acres (34.0 krn2 ). By April
13, the fire had consumed almost 20,000 acres (80.9 km 2 ) and was only 70%
contained. The ftre ultimately consumed over 26,000 acres (105.2 knl) of wilderness
area and was 90% contained by April 15. The total area bumed was increased by
back burning activities of the fire suppression crews. Back burning is used to
remove fuels ahead of the active fire front and impede the spread of the fire and
help Ill containment.
Hourly surface observations were used to provide an estimate of the passage of
the sea-breeze front. Unfiltered time series of temperature and relative humidity
data from four stations near the fire (not shown) indicate that Sanborn, the closest
of the four stations to the coast, shows the earliest peak in temperature ( 1,400 EDT),
\\bile the peak at the other stations is delayed by approximately 2 h. Following
the temperature maximum is a decrease in temperature spanning 6 - 8 h, which is
hardly tndicative of the passage of the sea-breeze front, and more closely resembles
the normal diurnal cycle. The unfiltered relative humidity time series does not
~hO\\ a marked change in air mass either. Using a high-pass filter to remove the
diurnal signal as described in the previous section reveals a stronger sea-breeze
ignal in both the temperature and relative humidity time series (Fig. 7 .2). It is
apparent that Sanborn shows the earliest start to humidity recovery around 1,700
EDT with the other stations following rather sharply an hour later.
The sea-breeze front can be seen clearly on the Tallahassee 0.5° elevation
base reflectivity imagery at 1828 UTC (Fig. 7.3). The East Fork Fire is visible
approximately 40 km southwest of the radar location and is indicated by a
ret1ectivity of approximately 18 d.BZ with the smoke plume traveling toward the
\Outheast (Fig. 7.3(a)). The sea-breeze front continues to move inland over the
ne:-.1 several hours and by 1927 UTC (Fig. 7.3(b)) the smoke plume from the frre
can be seen to increase in intensity with reflectivities reaching 28 - 32 dBZ for
a bnef period prior to the arrival of the sea-breeze front. By 2025 UTC the
)ca-breeze is just reaching the fire and the plume is displaying reflectivities in the
range of20- 25 dBZ near the frre (Fig. 7.3(c)), consistent with values observed
by Hufford et al. ( 1998) near the head of a rapidly moving wildfire. At 2124 UTC,
the sea-breeze front has reached the fire, as evidenced by the humidity recovery
tndicated by the Sanborn RAWS site. Despite the start of humidity recovery the
89
Remote Sensing and Modeling Applications to WildJand Fires
3
2
.."'....
~
0
~
- I
i§
5.
E
<>
t-
__.,_ Bloxham
->-Sanborn
- J - Tallahassee
~wilma
-2
-3
-4
-5
4/5 09
415 12
4/5 15
4/5 18
Time(EDT)
4/5 21
4/ 6 00
(a)
20
15
10
~
......
5
.Q
0
:.0
-E
::::>
.:::
....
I__.,_ Bloxham
--o-
-5
-o-
Sanborn
Tallahassee
-o- Wilma
-~ -10
u"'
~
- 15
-20
-25
4/5 09
4 /5 12
415 15
4'5 18
Time (EDT)
4/5 21
4/600
(b)
Figure 7.2 High-pass filtered time series of (a) temperature and (b) relative humidity
at the surface for the period starting at 0900 EDT 5 Apri I 2004 and ending at 0000
EDT 6 A pri I 2004, for several stations identified in Fig. 7 .I
plume intens ifies to levels not previous ly seen with a large core of the plume
now displaying reflectivities in the 28- 32+ dBZ range (Fig. 7.3(d)). Upper-level
winds at this time are carrying the plume back toward the coast, aided by the
upper-level return flow of the sea-breeze circulation. The reflectivity of the plume
decreases from this peak over the next seve ral hours as the fire begins to respond
to the decreased temperature and increased moisture of the marine air, leading to
a decline in fire intensity (Figs. 7.3(e) and (f)).
Base reflectivities from the 0.5 °. 1.5° and 2.5 ° elevations are used to examine
the vertical structure of the plume at 2,1 24 UTC. the time of greatest intensity
90
7
Interaction between a \Vildfire and the Sea-Breeze F ront
(c)
(e)
Figure 7.3
(d)
(f)
Radar reflectivity (dBZ, shaded as indicated) at a scan elevation of
o.s• at (a) 1828 UTC. (b) 1927 UTC, (c) 2025 UTC. (d) 2124 UTC, (c) 2223 UTC.
and (f) 2323 UTC for 5 April 2004
(Fig. 7.4). At the lowest elevation, the overall plume is 21.25 km in length with
Lhe region of reflectivity > 28 dBZ having a length of 8.75 km. The approximate
heights of the centroids of the ends of the plume at this elevation are 453 m and
638 m, well within the 800 m deep mixed layer (estimated from the 1200 UTC
sounding and observed high temperature). At the 1.5 elevation, the plume is
91
R em ote Sen sing a nd Modeling A pplications to Wildland F ires
slightly longer, 23.75 km (length of the 28+ dBZ core has shrunk to 3.75 km),
ranging in height from 1,151 - 1,637 m. At this e levation the plume is clearly
above the mixed layer and in the return flow of the sea-breeze circulation. The
2.5 ° e levation reveals a considerably smaller plume, 8.75 km in length (2.5 km for
the 28+ dBZ core), and spanning elevations of l ,849 - 2,157 m. The northwest
comers of the p lumes at each level are collocated, indicating that the p lume has a
rather strong vertical core immediately above the fue that would be consistent
with an intensely burning fue producing a strong buoyancy flux.
(c)
F ig ure 7.4 Radar reflectivity (dBZ. shaded as indicated) at 2124 UTC for 5 April
2004 at scan elevations of(a) 0.5°, (b) 1.5°, and (c) 2.5
92
7
Interaction between a Wildfire and the Sea-Breeze Front
The radial velocity field from the WSR-88D is a useful tool in identifying
areas of convergence and divergence. The 0.5° elevation radial velocity field from
Tallahassee shows a well defined convergence zone along the sea-breeze front as
it pushes inland against the prevailing northwesterly flow in the region (Fig. 7.5).
r\car the fire the pattern is more complex as the rising plume generates a region
of divergence oriented along the axis of the plume, possibly marking the return
flo"' of the thermal circulation generated by the intense heating of the fire. At
the 1.5 elevation, a divergence pattern is still discernible along the plume axis
although its extent is limited to the area near the peak in base reflectivity.
It should be noted that we have avoided making any direct comments as to the
magnitudes of the velocities shown. This is due to the relatively broad spectrum
width observed by the radar in the vicinity of the plume which is indicative of a
highly turbulent environment, and thus renders the magnitudes or the radar-derived
velocities less reliable.
(C)
Figure 7.5 Radial velocity (m·s- 1, shaded as indicated) at2,124 UTC for 5 April
2004 at scan elevations of (a) 0.5 ° and (c) 1.5°. Panels (b) and (d) provide close-up
representations of panels (a) and (c). respectively, with base reflectivities of 12 dBZ
and 28 dBZ indicated by the solid contours
93
Re m ote Sensing and Modelin g Applications to Wildland Fires
7.4
N umerical Simulations
In this Section, we present results from three numerical s imulations to explore
the nature of the interaction between a density current and a buoyant plume: two
simulations in which a density current and a plume arc examined in isolation,
respectively. and one in which both are present and arc allowed to interact.
The density current is initialized with a cold pool at one end of the domain
(Fig. 7 .6). When the simulation is started. the cold pool adjusts under gravity and
initiates a density current that travels in the positive y-direction. The ambient
winds arc directed in the negative y-direction, thus representing an offshore flow
situation. The cold pool is initially uniform in the x-dircction, but rapidly becomes
30 with the development of lobe and cleft instabilities (not s hown), consistent
with laboratory experiments and previous numerical investigations using threedimens ional models.
1.000
800
g
••
600
400
200
.------------.-----.----------------------..----------------- ------.--------...--______________
-------..-.-....--------------......
...... ..,_ .........................
......
-------------..--...--~---..----......
~-------.........................
......
...... ..-.- ......
...-.---..---~
..
--------------------..-..-------------·------------------ ·----------------....--------------..-.-..--..-.----------------..---------------------.----------------==:= =::: ::: ::: ::: ::: ::: ====: : =
--------------------------~-------
--------~~----- -------------------------~---------~....-----
~'C\~ ::3 := ::: ::: ::: ::: ::: ::: :::
== :::
-----------------------------
600
800 1.000 1.200 1.400 1.600 1.800 2.000 2.200 2.400 2.600 2.800 3.000 3.:!00
y (m)
2.5 m/s
Figu re 7.6 Model initial conditions in the y-= plane at x=480 m of potential
temperature (contour interval 1 K) and wind vectors in the plane of the section (in
m ·s 1 , vector scale shown at bottom right)
Figure 7.7(a) depicts the potential temperature, vertical velocity, and wind field
in the y- z plane. The Kelvin- Hernboltz billows characteristic of the interface
between the density current and the ambient atmosphere arc apparent, as is the
enhanced vertical motion and return flow associated with the density current.
Figure 7.7(b) illustrates the simulation of the plume only. The plume is initiated
by a heat source centered at y=2,700 m, the spatial configuration of which is a
smoothed top-hat function. Characteristic features of the p lume are similar to those
described in more detail in the simulations by Cunningham ct aJ. (2005).
The simulation in which both the plume and the density current are present is
depicted in Fig. 7.7(c). It is evident that the interaction between the density current
and the plume results in an apparent intensification of the plume, particularly in
the vertical velocity field, in conjunction with the arrival of the circulation
associated with the current, and that this intensification occurs before the head of
the density current reaches the heat source. The intensification apparently occurs
in response to the pressure perturbation that precedes the head of the current: this
94
7
Interaction b etween a Wildfire and the Sea-Breeze Front
pressure perturbation, which can impact the plume for a period of ti me before the
arrival of the colder, more dense air behind the head of the current, counteracts
lhe ambient flow at low levels su ch that the p lume becomes more vertical
(compare Fig. 7.7(b) with Fig. 7.7(c)), resulting in vertical velocities that are
significantly stronger than those seen in the plume in isolation.
1.~--------_--_--_---_---------------------_--_--_----------,
-------
800
~
__ ___ _____
---------
600
" 400
::
--------::
~~
-
=-=:-" r--:
-
-
..
-
-
- -- -- -- - - -------..... - - - - - -
...._
+-- - -
--
200
0
.~
800 1,000 1.200 1.400 1.600 1,800 2.000 2,200 2.400 2,600 2.800 3.000 3.200
y (m)
5. ili!s
~~::::::=:;;::::::=;=:;-:-::-:--:::-:-::-:--:­
-{).5-5.5-4.5-3.5-2.5-1.5-0.5 0.5 1.5 2.5 3.5 4.5 5.5 6.5
(a)
·-~
r--------------------:-.. . . . '. ---...., ....---------------------,
.
800
~
600
II
400
.§.
-
'
.
- - - -- "" - - - - - - - - - -- -- : :. - -
. ..' .-
- - '\
- - - -<>: :: ~·: -
200
:· ~ -
0~--~--~--~--~~~~--~--~--~--~--~-~-~
800 1.000 1.200 1.400 1,600 1.800 2.000 2.200 2.400 2,600 2.800 3,000 3.200
y {m)
5. ilils
-{).5-5.5-4.5-3.5-2.5-1.5~.5
0.5 1.5 2.5 3.5 4.5 5.5 6.5
(b)
1,000
r--------_--_-----:_-:.:,---=:--------_---_--_--_--_--_--_-----,
_____
=== =
:::: ::e ·::: ::__:: _
--~---..
--------- ___...... ..._ ..........
-~-.....;------
800
,..... 600
E
........
.......
·~-....
-""'-
~ 1-
~-
- - '""="'- ......
--- -- -....~·
" 400
---;-:::_, !(.
~
:
200~~~~~~~~~~~~~~~~~~_]j
~00 1,000 1.200 1.400 1.600 I ,800 2.000 2.200 2.400 2.600 2,800 3.000 3.200
Y (m)
..-.11!1~:::::::::;:::===;:::::::::::::::;~
-{).5-5.5-4.5-3.5-2.5-1.5~.5
5 m/s
0.5 1.5 2.5 3.5 4.5 5.5 6.5
(c)
Figure 7.7 Vertical velocity (in m·s- 1• shaded as indicated) and potential temperature
(contour interval I K) in the y-z plane for (a) the density current only, (b) the
plume only. and (c) the density current and the plume. Vectors depict velocity in
the plane of the section. with the scale in m ·s- 1 at the bottom right
95
Remote Sensing and Modeling Applications to Wildland Fires
7.5
Summary and Conclusions
Sea-breeze circulations are frequently observed to have a significant impact on
fire behavior; however, the nature of the interaction between wildfires and the
sea-breeze is poorly understood, and has not been studied extensively. In this
Chapter, radar observations of a plume associated with a wildfire in Florida were
presented that suggest that the arrival of a sea-breeze front results in a temporary.
but significant, increase in fire intensity. This intensification precedes the arrival
of the sea-breeze front at the location of the frre, such that the fire intensity is a
maximum at the time of, and slightly after, the passage of the front, and decreases
gradually thereafter with the arrival of cooler and moister air behind the front.
There is insufficient evidence to explain this intensification; however, the
idealized numerical simulations presented in Section 7.4 suggest that the period
of interaction preceding the arrival of the current may result in the temporary
intensification of vertical velocity in the plume. This increase in vertical velocity
appears to result from the interaction of the plume with a pressure perturbation
that precedes the head of the density current. The direct impact of this interaction
on fhe behavior is uncertain, however, and requires further study, particularly
using coupled atmosphere-fire models to explore the details of this interaction.
Another aspect of the wildland frre-sea-breeze system that warrants investigation
is how these interactions influence local and regional air quality.
Acknowledgements
The authors gratefully acknowledge Pam Brown of the USDA Forest Service in
Tallahassee for providing the RAWS data for this study, and Scott Taylor of the
Florida Division of Forestry for creating Fig. 7 .1. We also thank Dr. Rodman
Linn for insightful discussions on various aspects of the Chapter. DEH acknowledges
support from the Florida Department of Agriculture and Consumer Services. PC
acknowledges support from the USDA Forest Service, via a cooperative agreement
between the Florida State University and the USDA Forest Service Southern
Research Station, and from the FSU School of Computational Science, via a grant
of resources on the IDM SP Series 690 Power4-based supercomputer "Eclipse".
References
Atkins NT, Wakimoto RM , ( 1997), lnfluence of the synoptic-scale flow on sea breezes
observed during CaPE. Mon. Wea. Rev.. 125:2112-2130
Atkins NT, Wakimoto RM, Weckwerth TM, (1995), Observations of the sea-breeze front
during CaPE. Part Il: Dual-Doppler and aircraft analysis. Mon. Wea. Rev., 123: 944 - 969
96
7
Interaction between a \-Vildfire and the Sea-Breeze Front
Atlas D. ( 1960), Radar detection of the sea breeze. J Meteor.• 17: 244 - 258
Banta R\11, Olivier LD. l lolloway ET , Kropfli RA. Bartram BW, Crupp RE, Post MJ, ( 1992),
Smoke-column observations from two forest fires using Doppler lidar and Doppler radar. J.
Appl.•Heteor.• 3 1: 1328- 1349
Battan LJ, ( 1973). Radar Observation of the Atmosphere. University of Chicago Press. 324
Clark TL, Jenkms MA. Coen JL, Packham D R. (1996a). A coupled atmosphere-ftre model:
Convective feedback on fire line dynamaics. J. Appl. Meteor., 35: 875 - 901
Clark TL. Jenkins MA, Coen JL, Packham DR, (1996b), A coupled atmosphere-fire model: Role
of the convective Froude number and dynamic fingering a t the fire line. Int. J. Wildland
Fire, 6 : 177 - I 90
Clark TL., Coen JL, Latham D, (2004), D escription of a coupled atmosphere fire model. Jnt. J.
Wildland Fire, 13: 49 - 63
Colby FP, (2004). Simulation of the New E ngland Sea Breeze: T he effect of grid spacing. Wea.
Forecasting, 19 : 277 - 285
Cunningham P, Goodrick SL , Hussaini MY, Linn RR, (2005), Coherent vortical structures in
numerical simulations of buoyant p lumes from wild land fires. Int. J. Wildland Fire. 14:
61-75
Dailey PS, Fovell RG. ( 1999). Numerical simulation of the interaction between the sea-breeze
front and horizontal convective rolls. Part 1: Offshore ambienr flow. ,\tlon Wea. Rev.. 127:
858 - 878
Droegemeier KK, Wilhelmson RB, (1987), Numerical simulation of thunder.;tonn outflow
dynamics. Part 1: Outflow sensitivity experiments and turbulence dynamics. J . Atmos. Sci ..
44: 1180 - 1210
Fisher EL. ( 1960). An observational study of the sea breeze. J. Meteor., I 7: 645 660
Fleischmann CM. McGrattan KB, (1999), ~umerical and experimental gravity currents related
to backdrafts. Fire Safety J., 33 : 21 -34
Fovell RG. (2005), Convective initiation ahead of the sea-breeze front. Mon. Wea . Rev.. 133:
264-278
Fovell RG, Da1ley PS. (200 1). Numerical simulation o f the interaction between the sea-breeze
front and horizontal convective rolls. Part II: Alongshore ambient flo~. Mon. Wea. Rev. ,
129: 2057 - 2072
I !iirtel C, Meiburg E. Necker F, (2000a), Analysis and direct numerical simulation of the flow at
a gravity-current head. Part l. Flow topology and front speed for slip and no-slip boundaries.
J Fluid Mech., 418: 189 - 212
Hartel C. Carlsson F. Thunblom M. (2000b). Analysis and djrect numerical simulation of the
flow at a grav1ty-current head. Part 2. The lobe-and-cleft instability. J. Fluid Mech .. 418:
2 13 - 229
Hufford GL, Kelley I fL, Sparkman W. Moore RK. ( 1998). Use of real-time multisatellite and
radar data to support forest fire managemenL Wea. Forecasting, 13: 592 - 605
Linn RR. (1997). A transport model for prediction of wildfire behaviOr. Scientific Report
LA-13334-T, Los A lamos National Laboratory, 195
Linn RR, Cunningham P, (2005). Numerical simulations of grass fires using a coupled
atmosphere- fire model: Basic fire behavior and dependence on wind speed. J . Geophys.
Res., in press
97
Remote Sens ing and Modeling Applications to Wildland F ires
Linn RR, Reisner JM, Colman JJ. Winterkamp J, (2002), Studying wildfire behavior using
FIRETEC. Int. J. Wildland Fire. 11 : 233- 246
6zgokrnen TM. Fisher PF. Dian J. Iliescu T, (2004), Three-dimensional turbulent bottom density
currents from a high-order nonhydrostatic spectral element model. J . Phvs. Oceanogr.. 34:
2006-2026
Pielke RA. ( 1974). A three-dunensionaJ numerical model of the sea breezes over south Florida.
Won. Wea. Rev., 102: 115- 139
Rao PA, Fuelberg HE, Droegemeier KK, (1999), High-resolution modeling of the Cape
Canaveral area land- water circulations and associated features. Mon. Wea. Rev., 127:
1808 - 1821
Rogers RR, Brown WOJ, ( L997). Radar observations of a major industrial fire. Bull. Amer.
Meteor. Soc., 78: 803 - 814
Sha W, Kawamura T, Ueda H. (1991), A numerical study on sea/land breezes as a gravity
current: Kelvin- Helmholtz billows and inland penetration of the sea-breeze front. J. Atmos.
Sci., 48: 1649- 1665
Simpson JE, (1994), Sea Breeze and Local Winds. Cambridge University Press, 234
Simpson JE, Mansfield DA, Milford JR. (1977). Inland penetration of sea-breeze fronts. Quart.
J Roy Meteor. Soc., 103: 47-76
Skamarock WC. Klemp JB, Dudhia J. (2001), Prototypes for the WRF (Weather Research and
Forecasting) model. Preprints. Ninth Conf. On Mesoscale Processes, Fort Lauderdale, FL.
A mer
lvleteor Soc., J 1 I - 115
Wakimoto RM. ( 1982), The life cycle of thunderstorm gust fronts as viewed with Doppler
radar and rawinsonde data. Mon. Wea. Rev., 110: I 060- I 082
Wakimoto RM, Atkins '\IT, (1994). Observations of the sea-breeze front during CaPE. Part I:
Single-Doppler. sateiJite, and cloud photogramrnetty analysis. Mon. Wea. Rev.. 122:
1092 1114
Wilson JW, Weckwerth TM, Vivekanandan J, Wakimoto Ra\11, Russell RW, {1994), Boundary
layer clear-air radar echoes: Origin of echoes and accuracy of derived winds. J. Atmos.
Oceanic. Techno/., 11: 1184-1206
Xu Q. Xue M, Droegemeier KK, ( 1996). Numerical simulations of density currents in sheared
environments within a vertically confmed channel. J. Atmos. Sci.. 53: 770 -786
Xuc M, Xu Q, Droegemeier KK, (1997), A theoretical and numerica l study of density currents
in nonconstant shear flows. J. Atmos. Sci., 54: 1998 - 2019
98